XTIDE tech support thread

[hargle]

I think i know what is the problem with flashing.

I assembled the two new XTIDE kits. Both of the cards worked on first try except that i couldn't flash the other with SDP enabled.

The other card had Atmel AT28C64X instead of AT28C64B. I replaced the X EEPROM with another B EEPROM and flashing was successful. The Atmel EEPROM ending with X does not appear to support Software Data Protection.

are you serious? we got a mix of parts from mouser? I've shipped everything out now, so I can't go check what I have/had.

k2x4b524[, can you please check your atmel part numbers and post back here? I've got a serious bone to pick with mouser if they did this to us. Still better than jameco, but man, that is seriously lame. I suspect there is no way to detect via software which part is installed in the system to automaticaly know to turn on SDP or not.
uuuugh.

 

[hargle]

Attached a seagate ST364218 6.4 Gig, formatted with win98 ms-dos, swapped
back onto xt-ide card
Boot sector found...but froze.

Are you sure win98 DOS works on an 8088 machine? I suspect they would be doing some 386 instructions.

Stick with DOS 6.2 or less.

....xt-ide failed to detect drive, reboot, found drive

Until you get this solved, there is no point in working with any operating system. the card *must* detect the drive every single time it boots.

This comes down to:
1) bad drive or cable
2) IO resource conflict
3) bad ICs
4) bad soldering job

I assume you've swapped out #1 a few times.
for #2, remove all other cards from the machine.
#3 is probably not an issue. you may have a bent leg on an IC though.
#4 is, unfortunately, the most likely culprit. Assuming you built this card yourself. If I built it for you, then we should maybe do a card exchange so I can diagnose it.

You have it very close to working, so I suspect a cold solder joint or two is keeping it from working 100%.

If the BIOS is coming up on the screen every time you boot, then the problem is in the IO system. According to Andrew, all of the ICs except U9 and U10 are used in the IO system, so I would first try and re-flow all the solder on all the sockets except for those 2. All you have to do is push the iron against the socket pin until it moves slightly. Adding more solder may or may not help. If your solder ends up sticking to the iron and forms icicle like formations when you pull the iron away, then it is not hot enough, and it is not making a good connection.

 

[NobodyIsHere]

Hi! Regarding soldering, less is definitely better. Let your soldering iron warm up so you can tin the tip with some solder Then use the cleaned off soldering iron to remove as much excess solder as possible. Even use solder wick to pull away as much as you can. I hold the board on its side and let the melted solder "drain" on to the soldering iron to remove excess solder.

From the description, it sounds like a cold joint someplace. The first suspects are VCC and GND pins so I would look there first. Make sure the sockets are good and clear too by removing and replacing the chips a few times. Make sure there are no bent under pins.

I hope this helps. Thanks and have a nice day!

Andrew Lynch

 

[k2x4b524[]

Hargle, all my bits are B series, cept 1 which was an X series, was with the kit, flashed for tandy bios, i gave the kit, and 3 chips to him for modding to his pc/jr, one of those WAS an X, but took the flash properly. I do have a question though.
Whats the difference between xt_ide.bin and xtp_ide.bin when i flash the plus one, it fails, same with the at, but xt ide fits just right, i think that may be the computer i'm using so no biggie.

But hargle, one of my chips WAS definitely an X series

 

[Lord Moz]

aitotat can explain better, but the XT, XT+ and AT versions all have optimizations for supporting successively higher class machines.
The XT+ requires a V20/V30 or better and will not work on an 8088 or 8086. The AT version requires a 286.

Hargle: If you are interested, all 3 of the Atmel parts you sent with my package are "B" models.

__
Trevor

 

[k2x4b524[]

that said, could i take an xt-ide flashed with the at code to a tandy 1000-RLX?

 

[carlsson]

FWIW, both my two kits have the intended AT28C64B part.

 

[salamontagne]

Fu**k!!

Fu**k!!

Well, i totally blew it.

After getting the card to boot ms-dos 6 from the seagate i listed below, i decided to try to do a format after booting the system off of a floppy.

I booted it again, then went todo a re-name of config.sys and autoexec. after renameing the
files from the disk it booted from, I got a general failure.

I think ive got the failue traced down to the floppy controller i'm useing. While the Winbond "should" be backwards compatiable in an 8-bit slot, and hasnt cause an issue so far, its probabily the culprit when useing the xt-ide to boot from it.

Unfortunetly, I was unable to get the MFM controller (Seagate st11 bios revision 2.1) to play nice
with the xt-ide. Tried changeing the xt-ide's I/O to 280, and a few differant memory addresses, and tried the seagate controller in all if its known arrangements, no dice.

While giveing the xt-ide a good once over, i decided to straighten out a few pins a bit better...

I really should be building bicycle's..... I broke one of the pins clean off, the one right above the one i snipped before. I must have totally overstressed the ones i bent out of the way.

Would anybody care to replace the header for me? Just PM me with your address and how much you want to solder in a new one....

 

[aitotat]

Whats the difference between xt_ide.bin and xtp_ide.bin

ide_xt.bin (XT build)

• Work on any x86 CPU since only 8088/8086 instructions are used
• Slowest transfer rates since INS and OUTS instructions are not supported

ide_xtp.bin (XT+ build)

• Requires 80188/80186, NEC V20/V30 or better
• Faster transfer rates and some minor optimizations

ide_at.bin (AT build)

• Meant for AT systems with 286 or better and 16-bit ISA (CPU requirements are the same as XT+ build)
• True 16-bit AT-IDE transfers supported (i'll support them on other builds for next version)
• OS Hooks to notify operating system that seek is in progress and other processing can be done

I'll include some sort of readme for next version so everyone can select best build and settings. I have posted some transfer rate comparisons with XT and XT+ build on the XTIDE Universal BIOS thread.

 

[hargle]

While giveing the xt-ide a good once over, i decided to straighten out a few pins a bit better...

I really should be building bicycle's..... I broke one of the pins clean off, the one right above the one i snipped before. I must have totally overstressed the ones i bent out of the way.

I think you got lucky. I believe that pin is ground, and is probably not needed, since there are plenty of other grounds nearby.
Replacing the header would be quite a bit of work with desoldering and possibly ripping up traces from the board. Have you tried it without that connector?

 

[k2x4b524[]

Success, it works in the 1000TL, Even with my 40mb ST351 A/X, But a Note to Tandy 286 users, Set the system speed to slow, otherwise, the tandy will go through it's boot sequence BEFORE the hard drive finishes initializing, also, beware of dos 5.0 some hard drives it will boot and some it wont, but 6.22 boots from them all, and if the partition is small enough dos 2 thru 4 will aswell, IBM dos performs better than MS dos, Tandy dos hates it.... Will try free dos soon as i get a fully working os install tested.

 

[aitotat]

Success, it works in the 1000TL, Even with my 40mb ST351 A/X, But a Note to Tandy 286 users, Set the system speed to slow, otherwise, the tandy will go through it's boot sequence BEFORE the hard drive finishes initializing

Can you be more specific? There should be no need to change system speed.

beware of dos 5.0 some hard drives it will boot and some it wont

What drives it won't boot? I haven't had any problems with DOS 5.

IBM dos performs better than MS dos, Tandy dos hates it

IBM DOS 5 should be just a rebranded version of MS-DOS so there shouldn't be any differences. Tandy DOS 3.2 does not support large drives. Do you have problems with Tandy DOS 3.3?

 

[k2x4b524[]

it refuses to boot my quantum 52mb and my seagate 332mb 5.25 ide hard drive. I haven't had much luck with dos 5 for some reason, it always hangs just AFTER it sees the hard drive, the other versions dont.

I mean the Tandy does it's post and initializes the XT-IDE bios before the hard drive finishes spinning up or doing it's head check, when it's set at normal speed. when the Tandy is at slow speed, the hard drive gets a chance to finish spinning up BEFORE the XT_IDE initializes.

Aitotat, whens rc 3 coming out?

 

[aitotat]

it refuses to boot my quantum 52mb and my seagate 332mb 5.25 ide hard drive.

And both of them boot with DOS 6? I don't have those drives but i have 40MB Quantum. I'll try with it. What was the exact version of DOS 5 that you tried?

I mean the Tandy does it's post and initializes the XT-IDE bios before the hard drive finishes spinning up or doing it's head check, when it's set at normal speed.

Let me guess: Tandy doesn't do any RAM testing? A very fast POST might cause the thing you described. I use very short timeout values during drive detection to minimize delay when master or slave drive is not present. I'll add a short (configurable) delay before drive detection starts. It should help and it won't slow down detection when there are many controllers with few drives.

Aitotat, whens rc 3 coming out?

After two or three weeks maybe.

 

[k2x4b524[]

plain jane MS-DOS 5, i might be doing something wrong, but my quantum wont take dos 5, dos 6.22 is a champ.
And Nope, the tandy doesn't do memory testing, just shows the whole amount, in this case, 786kb, hence the slow down from fast to slow as it breezes by really quick.

I am champing at the prospect of HD floppy support through xt-ide, i've got a 4-floppy drive card, that might take the xt-ide bios, if it can, the hd floppy code sits right in. Aito, if you do that? how are you going to bypass the XT's built in drive seek? there is no dipswitch for 0 floppy drives

 

[aitotat]

I installed MS-DOS 5.00 to a 40MB Quantum ProDrive 40AT. I installed it with the DOS install program and installing* and booting worked just fine.

About HD floppy support. There is not enough free space at the moment. Boot menu with necessary libraries take about half of what is used at the moment. There are enough free space for EBIOS functions (support for drives over 8.4GB) since most of the existing IDE related code can be used with EBIOS functions. How large is the BIOS on HD floppy controllers?

Floppy support from main BIOS won't be a problem since HD floppy BIOS simply intercepts and handles all floppy function calls.

*keyb.com won't work on my clone XT so i can't use international keyboard layouts. I also had to remove keyb.com from autoexec.bat on the DOS install disk.

 

[k2x4b524[]

could a bios revision without the boot menu be served up? for those of us that might want to give it a shot. Just set a default to scan the floppies first, then the hdd. since it already autodetects the hard drive, that shouldn't be a problem, also would it be possible to piggy back chips, like they did in the AT for the ram?

 

[salamontagne]

Success! well, mostly

Success! well, mostly

I re-installed the Xt-ide adapter along with the Seagate ST36421A. XT-IDE Finds the drive everytime, no hassles. For the record, this is with the extra pin accidently snapped off and useing a short 2 drive 40-conductor cable. I'm also useing ide_xt.bin RC2 and dos 5.0. Default settings on the XT-IDE controller.

The only prolbem I'm haveing now is with a mfm controller installed in the system (Rev 1.7 or 2.1 of bios), It refuses to boot from the XT-IDE. I'm not exactally sure why at this point. Ive tried turning the drive translation on and off, but i still get he "Non-system disk or disk error" while trying to boot from the ide drive with the mfm controller installed.

Changeing around the MFM drive controllers location (in memory, useing the jumper block) dosen't have any effect other then preventing XT-IDE's bios from popping up.

Since many posts in the forum point to dos 5 being prolbematic, I'll say its probabily a dos 5 issue. The MFM drive (which needed a new control board) is on its last legs anyhow, so I'm going to request my friend use what time he has left to get the software he wants on the machine finailized and from there, we'll probabily use a IDE-CF adapter with the write protection tab enabled and go from there.

A big thanks to hargle and aitotat for makeing this all happen!

EDIT: For those useing an XT-IDE and an MFM controller, REMOVE the XT-IDE board if your going to LLF the MFM drive useing its own
bios or else the system will re-boot when you go into entering the defects map.

 

[hargle]

Changeing around the MFM drive controllers location (in memory, useing the jumper block) dosen't have any effect other then preventing XT-IDE's bios from popping up.

This to me sounds like you managed to put the XTIDE BIOS on top of the MFM's BIOS or some other conflict in address decoding. There is absolutely no reason that a system shouldn't be able to load as many option roms as you have installed in the machine.

EDIT: For those useing an XT-IDE and an MFM controller, REMOVE the XT-IDE board if your going to LLF the MFM drive useing its own
bios or else the system will re-boot when you go into entering the defects map.

That too tells me that you've got an addressing conflict.

 

[Marrr]

So I finally did some testing, as of yet only with the following:

XT clone, CXQ70108D-8 @ 8MHz
WDAC22100-23H (2 GB)
IBM DOS 5.02, CheckIt 3.0, and some other software

Issues so far:
- Never detects the drive after power-on, always detects after reset, I'm pretty sure it's because this drive is slow to spin-up. As far as I know it's going to be fixed in next BIOS version, isn't it?
- On my modern PC this HDD is seen as (C/H/S) 1023/64/63, XT-IDE however detects it as 1022/64/63. So, if partitioned on the modern PC, data on the last cyllinder is inaccessible. Is this a bug, or users aren't supposed to partition&format drives with alien controllers/BIOSes?
- 300h is default address for many (all?) ethernet cards, consider using another default in the future. (added a warning into Wiki)
- How many I/O ports XT-IDE actually occupies? My groping with debug made me suspect it's the range BASE..BASE+F, ie. 16 ports, right?
- Wiki claims default is JP2=Off (Disables writing to the ROM). Not anymore, right?
- XT-IDE steals 1KB of RAM at 639KB address, I guess for its BIOS' variables? Are there chances for a more elegant solution?

Other comments:
- Benchmarking with CheckIt gives 176.1 KB/s (XT) or 212.0 KB/s (XT+), which already is faster than ST-225 on the same machine - 133.4 KB/s. (added to Wiki)
- Interesting thing: seek times on XT with XT-IDE are slightly shorter than on a modern PC: 9.7 and 1.8ms, compared to 10.0 and 2.1ms. Probably thanks to no-IRQ?
- Absolutely no problems with flashing, all my Atmels are 'B' variant.

Overall, as of yet I'm happy, and expecting to be very happy once the spin-up issue is resolved...